Convergence in distribution of sums of bivariate Appell polynomials with long-range dependence (Q1175664)

Let \(\{\xi_ k, k\in \mathbb{Z}\}\) be a sequence of independent identically distributed random variables with \(E \xi_ 0=0\), \(E \xi_ 0^ 2=\sigma^ 2<\infty\), and \(X_ j=\sum_{s=-\infty}^ \infty b_{j- s}\xi_ s\) be a random sequence of moving average. The coefficients \(b_ k\) are defined by \[ b_ k=\int_{-\pi}^ \pi e^{-ikx}| x|^{-\alpha/2}L_ 1^{1/2}(| x|^{-1})dx, \] where \(L_ 1\) is slowly varying at \(\infty\). Let \(P_{m,n}(x,y)\) be a bivariate Appell polynomial of \(m,n\) degree. Limit theorems for normalized forms of the type \(S_ N=\sum_{k=1}^ N a_{j-k}P_{m,n}(X_ j,X_ k)\) are proved. \(a_ k\) is defined by \[ a_ k=\int_{-\pi}^ \pi e^{- ikx}| x|^{-\beta}L_ 2(| x|^{-1})dx \] with \(L_ 2\) being slowly varying. The limit distribution of \(S_ N\) as \(N\to\infty\) is proved to be Gaussian or it is distributed as multiple Itô-Wiener integral. The type of limit distribution depends on \(\alpha,\beta,m,n\) and on some relations between these parameters. The techniques used by \textit{L. Giraitis} and \textit{D. Surgailis} [Lith. Math. J. 29, No. 2, 128- 145 (1989); translation from Litov. Mat. Sb. 29, No. 2, 290-311 (1989; Zbl 0689.60031)] for sums of univariate Appell polynomials are extended.